AI Paper Atlas

Research map / Video / Video-language understanding

FrameThinker: Learning to Think with Long Videos via Multi-Turn Frame Spotlighting

Zefeng He, Xiaoye Qu, Yafu Li, Siyuan Huang, Daizong Liu, Yu Cheng

ICLR 2026 · Video › Video-language understanding · approach: Long video understanding

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace uniform 32-frame sampling with multi-turn frame spotlightingQwen2.5-VL-7B · LongVideo-Reasonaccuracy (%): 64.1 -> 76.1Improved
train with FrameThinker (SFT+RL) on Video-HolmesQwen2.5-VL-7B · Video-Holmesaccuracy (%): 27.8 -> 56.1Improved
large-scale FrameThinker trainingQwen2.5-VL-7B · six benchmarks averageaccuracy (%): +10.4Improved
FrameThinker versus Video-R1 on MLVUVideo-R1 · MLVUaccuracy (%): 60.2 -> 59.1Hurt
add a format reward (ablation)FrameThinker RL · Video-Holmesaction usageHurt
use unconditional action bonus instead of conditional (ablation)FrameThinker RL · Video-Holmestraining stabilityHurt
add Cognitive Consistency Verification (CCV) in training and inference (ablation)FrameThinker · Video-HolmesaccuracyImproved
larger bonus for get frame number than choose frames (0.2 vs 0) (ablation)FrameThinker RL · Video-HolmesaccuracyImproved
reward more turns, R_action = k(T-1) (ablation)FrameThinker RL · Video-Holmestraining stabilityHurt
apply FrameThinker SFT/RL setup to baseline Qwen2.5-VL-7B versus full approach (ablation)FrameThinker · Video-HolmesaccuracyImproved

Later papers that compared with this method

2 later accepted papers report FrameThinker as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy